Everything about DeepSeek V4 Pro 0813

What you need to know

DeepSeek V4 Pro 0813 is a language model from DeepSeek, released on August 13, 2026. That makes it 2 months old as of October 22, 2026. It is an open-weight model: you can host it yourself on infrastructure you control, and 43 API providers listed on models.dev also offer it. It has a reasoning mode: it can "think" before answering, which helps with multi-step problems but uses more tokens, and therefore more money.

We file it under open-weight models (68 models in this category on the site).

What do independent benchmarks say?

According to results published by Epoch AI (checked on October 22, 2026), DeepSeek V4 Pro 0813 has a capability index of 72.2 out of 100, computed on 5 benchmarks from our basket.

It is not among the top 30 of the weekly ranking.

• GPQA Diamond (PhD-level science questions): 91.7% — the best tracked model, Gemini 4 Argon, reaches 96.4%
• OTIS Mock AIME (competition mathematics problems): 98.6% — the best tracked model, GPT-5.5 Pro, reaches 100%
• FrontierMath (very hard research-level mathematics): 64.6% — the best tracked model, GPT-6 Astra, reaches 93.7%
• FrontierCode (advanced programming tasks): 28.6% — the best tracked model, Claude Opus 5.5, reaches 54.6%

What it costs

DeepSeek V4 Pro 0813 costs $0.66 per million input tokens and $1.98 per million output tokens (developer's price, checked on October 22, 2026).

In practice, an agent task that reads 20,000 tokens and writes 2,000 costs about $0.017, or $17.16 for 1,000 such tasks.

That is 2 times the median output price of open-weight models ($1).

Tokens read back from cache cost $0.022 per million, which helps an agent that keeps resending the same context.

To estimate your own budget, use the agent cost calculator.

Technical specifications

Its context window reaches 1,048,576 tokens, roughly 1,573 pages of text at 500 words per page: enough to load a whole book or a mid-sized codebase. A single answer can run to 393,216 tokens.

It accepts as input: text.

It can call tools (functions, APIs) and return structured JSON output, which makes it a possible foundation for an AI agent.

The weights are published on Hugging Face (deepseek-ai/DeepSeek-V4-Pro-0813). See also our list of open-source models.

What is it good for?

Given its category, DeepSeek V4 Pro 0813 is mainly aimed at projects where you want control over your data or hosting, or want to adapt the model to your field.

Its very large context window makes it worth considering for analysing large sets of documents in a single request.

Before adopting it, test it on your own examples: benchmarks measure standard tasks, not your job.

What could replace it?

In the same category, the models closest to DeepSeek V4 Pro 0813 are:

• Qwen3.6 27B (Alibaba Cloud (Qwen), index 72.2) — see the comparison
• GLM-5.3-Flash (Zhipu AI (Z.ai), index 71.9) — see the comparison
• GLM-5.1 (Zhipu AI (Z.ai), index 70.9) — see the comparison
• Kimi K2 Thinking (Moonshot AI, index 69.6)